A comparison of multimodal therapy and
surgery for esophageal adenocarcinoma.
Item Type Article
Authors Walsh, T N;Noonan, N;Hollywood, D;Kelly, A;Keeling,
N;Hennessy, T P
Citation A comparison of multimodal therapy and surgery for esophageal
adenocarcinoma. 1996, 335 (7):462-7 N. Engl. J. Med.
DOI 10.1056/NEJM199608153350702
Journal The New England journal of medicine
Rights Archived with thanks to The New England journal of medicine
Download date 09/08/2022 21:41:14
Link to Item http://hdl.handle.net/10147/621423
Find this and similar works at - http://www.lenus.ie/hse
462
August 15, 1996
The New England Journal o
f
Medicine
A COMPARISON OF MULTIMODAL THERAPY AND SURGERY
FOR ESOPHAGEAL ADENOCARCINOMA
T
HOMAS
N. W
ALSH
, M.D., N
OIRIN
N
OONAN
, M.B., D
ONAL
H
OLLYWOOD
, P
H
.D., A
LAN
K
ELLY
, P
H
.D., C.S
TAT
.,
N
APOLEON
K
EELING
, M.D.,
AND
T
HOMAS
P.J. H
ENNESSY
, M.D.
A
BSTRACT
Background
Uncontrolled studies suggest that a
combination of chemotherapy and radiotherapy im-
proves the survival of patients with esophageal ade-
nocarcinoma. We conducted a prospective, random-
ized trial comparing surgery alone with combined
chemotherapy, radiotherapy, and surgery.
Methods
Patients assigned to multimodal therapy
received two courses of chemotherapy in weeks
1 and 6 (fluorouracil, 15 mg per kilogram of body
weight daily for five days, and cisplatin, 75 mg per
square meter of body-surface area on day 7) and a
course of radiotherapy (40 Gy, administered in 15
fractions over a three-week period, beginning con-
currently with the first course of chemotherapy), fol-
lowed by surgery. The patients assigned to surgery
had no preoperative therapy.
Results
Of the 58 patients assigned to multimodal
therapy and the 55 assigned to surgery, 10 and 1, re-
spectively, were withdrawn for protocol violations.
At the time of surgery, 23 of 55 patients (42 percent)
treated with preoperative multimodal therapy who
could be evaluated had positive nodes or metasta-
ses, as compared with 45 of the 55 patients (82 per-
cent) who underwent surgery alone (P
0.001). Thir-
teen of the 52 patients (25 percent) who underwent
surgery after multimodal therapy had complete re-
sponses, as determined pathologically. The median
survival of patients assigned to multimodal therapy
was 16 months, as compared with 11 months for
those assigned to surgery alone (P
0.01). At one,
two, and three years, 52, 37, and 32 percent, respec-
tively, of patients assigned to multimodal therapy
were alive, as compared with 44, 26, and 6 percent
of those assigned to surgery, with the survival ad-
vantage favoring multimodal therapy reaching sig-
nificance at three years (P
0.01).
Conclusions
Multimodal treatment is superior to
surgery alone for patients with resectable adenocar-
cinoma of the esophagus. (N Engl J Med 1996;335:
462-7.)
©1996, Massachusetts Medical Society.
From the Departments of Surgery (T.N.W., T.P.J.H.) and Gastroenterol-
ogy (N.N., N.K.), St. James’s Hospital; the Department of Radiotherapy,
St. Luke’s Hospital (D.H.); and the Departments of Community Health
and Statistics, Trinity College (A.K.) — all in Dublin, Ireland. Address re-
print requests to Dr. Walsh at the Department of Surgery, Beaumont Hos-
pital, Dublin 9, Ireland.
URING the past 20 years the rate of
increase in the incidence of adenocar-
cinoma of the esophagus has outstripped
that of all other tumors.
1-4
This dramatic
change has been accompanied by a shift in the bio-
logic behavior of esophageal cancer toward poorer
differentiation and greater nodal involvement.
5
At
D
the time of resection 85 to 95 percent of patients
have lymph-node involvement, and after standard
surgical resection fewer than 10 percent survive five
years.
5,6
The role of more radical surgery is contro-
versial, and no randomized trials have been reported.
Since most patients have systemic disease at presen-
tation,
7,8
systemic therapy may improve the outcome.
Single-drug or multidrug chemotherapy rarely in-
duces a complete response, as determined patholog-
ically, and does not enhance survival.
9
When pa-
tients with potentially curable cancer were treated
with radiotherapy alone, the two-year survival was
only 10 percent.
10
Neoadjuvant therapy given before
surgery may reduce the incidence of micrometasta-
ses, increase resectability, control systemic disease,
and allow accurate assessment of the completeness
of the pathological response, all of which might
influence decisions on postoperative treatment. A
number of studies have compared the response of
patients to neoadjuvant therapy with the outcome in
historical control patients
11-20
and have shown that
esophageal adenocarcinoma responds to treatment
based on fluorouracil and radiotherapy. We report a
trial in which a combination of preoperative chemo-
therapy and radiotherapy was compared with sur-
gery alone.
METHODS
Patients
In May 1990, we undertook a randomized, controlled trial to
compare the outcome of multimodal treatment, consisting of two
courses of fluorouracil and cisplatin and 40 Gy of radiotherapy fol-
lowed by surgery, with the outcome of surgery alone for esopha-
geal adenocarcinoma. Patients who met any of the following crite-
ria were excluded from the study: age greater than 76 years, distant
metastases, carcinoma of the cervical esophagus requiring laryn-
gectomy, leukocyte count of less than 3500 per cubic millimeter,
platelet count of less than 100,000 per cubic millimeter, serum cre-
atinine concentration above 1.4 mg per deciliter (124
m
mol per li-
ter), an Eastern Cooperative Oncology Group (ECOG) perform-
ance status of 3 or 4, previous chemotherapy or radiotherapy,
previous malignant condition (apart from skin cancer), coexisting
disease contraindicating surgery, and social circumstances not con-
ducive to compliance with the full treatment protocol. The study
The New England Journal of Medicine
Downloaded from nejm.org at ST LUKES INST CANCER RSCH on June 8, 2017. For personal use only. No other uses without permission.
Copyright © 1996 Massachusetts Medical Society. All rights reserved.
COMPARISON OF MULTIMODAL THERAPY AND SURGERY FOR ESOPHAGEAL ADENOCARCINOMA
Volume 335 Number 7
463
was approved by the St. James’s Hospital Ethics Committee, and
informed consent was obtained from all patients.
Preoperative Tumor Staging
The extent of the tumor was evaluated in each patient by phys-
ical examination, chest radiography, abdominal ultrasonography,
and upper gastrointestinal endoscopy. Bronchoscopy was per-
formed when indicated by symptoms, the location of the tumor,
or chest radiography. Computed tomography was performed in
selected patients with equivocal findings on chest radiographs or
liver ultrasonograms. Isotope bone scans were occasionally per-
formed if indicated.
Patient Monitoring
Initially the patients were hospitalized for the duration of treat-
ment, but later in the study they were routinely admitted only for
chemotherapy. Many who could commute to the radiotherapy fa-
cility received radiotherapy as outpatients. Patients were inter-
viewed and examined daily during chemotherapy. The following
hematologic and biochemical studies were performed at least
twice weekly: platelet and leukocyte counts and measurements of
hemoglobin and serum electrolytes, creatinine, bilirubin, alkaline
phosphatase, and
g
-glutamyltransferase. Arterial-blood gas analy-
sis and pulmonary-function tests were performed at the outset
and repeated when indicated.
Chemotherapy
Chemotherapy consisted of two courses of fluorouracil and cis-
platin. Fluorouracil (15 mg per kilogram of body weight per day)
was infused over a period of 16 hours on days 1 through 5. After
one day of hydration with 2 liters of 0.9 percent saline, cisplatin
(75 mg per square meter of body-surface area) was infused over a
period of eight hours on day 7. This cycle was repeated in week 6.
Radiotherapy
Radiotherapy was begun on the first day of the first course of
chemotherapy and given for a total of 15 days (days 1 to 5, 8 to
12, and 15 to 19). All patients were treated with megavoltage-
therapy units with either 4-MV photons (Cobalt model SEM100,
Fairy Engineering, or Phillips model SL75-5) or 8-MV photons
(Dynaray model 10, Radiation Dynamics). The treated area ex-
tended 5 cm beyond the longitudinal margins of the tumor, as
defined by endoscopic and radiologic examination, and 2 to 3 cm
beyond the radial margins. Initially, all patients were treated with
parallel opposed fields (anteroposterior and posteroanterior). The
technique was modified in 1994 to a three-field arrangement (an-
terior, right-posterior, and left-posterior oblique fields) to dimin-
ish the amount of radiation to the spinal cord. With the parallel
opposed fields, a midline dose of 40 Gy in 15 fractions was pre-
scribed (2.67 Gy per fraction). With the three-field technique, a
dose of 40 Gy
10 percent in 15 fractions was delivered to the
entire treatment volume (2.67 Gy per fraction) with a computer-
ized treatment-planning system (AECL/Theratronics Therplan).
There was no correction for transmission of radiation to the lungs
during either treatment technique.
Surgery
Surgery was carried out eight weeks after the beginning of
treatment but was delayed if the leukocyte count was less than
2500 per cubic millimeter or the platelet count was less than
100,000 per cubic millimeter. Five operative approaches were
used. Tumors of the cardia were resected through an approach in-
volving the abdomen and the left side of the chest. In selected
patients total gastrectomy and distal esophagectomy were per-
formed through an abdominal approach. Tumors of the lower
third of the esophagus were resected by the Lewis–Tanner oper-
ation, and tumors of the middle third by a three-stage operation
in which the esophagus was mobilized in the right side of the
chest and the anastomosis performed in the neck. In patients with
poor respiratory reserve the transhiatal approach was used. Intes-
tinal continuity was restored by placing the stomach in the pos-
terior mediastinum. A single layer of interrupted linen sutures was
used for the anastomosis. Patients were extubated in the operat-
ing room or the recovery room, returned to the intensive care
unit for four days on average, and were usually discharged less
than three weeks after surgery.
Treatment-Induced Toxicity
A toxic reaction was defined according to the criteria of the
World Health Organization (WHO).
21
With the introduction of
ondansetron early in the study for the treatment of cisplatin-
induced nausea, all patients were treated prophylactically before
and during the cisplatin infusion. If the leukocyte count fell be-
low 2500 per cubic millimeter or the platelet count fell below
100,000 per cubic millimeter, chemotherapy was withheld or ra-
diotherapy or surgery was delayed until the count recovered. An
interruption in the treatment regimen of more than two weeks
was considered a major deviation from the protocol.
Pathological Stage
The tumor stage was defined according to the classification of
the American Joint Committee on Cancer.
22
The stage of the can-
cer was defined in each patient according to the location and ex-
tent of any residual disease after chemotherapy and radiotherapy.
The absence of residual tumor in the resected specimen, includ-
ing the lymph nodes, was defined as a complete pathological re-
sponse (stage 0). The tumor was classified as stage 1 if there was
residual tumor in the mucosa or submucosa and the lymph nodes
were free of tumor. If residual deposits involved the muscularis
propria or adventitia but the lymph nodes were free of tumor, the
disease was classified as stage 2a. If there was no residual tumor
in the esophagus but a tumor deposit was found in the nodes, it
was classified as stage 2b. If tumor breached the esophageal wall
and the lymph nodes also contained tumor, it was defined as stage
3. Stage 4 disease was defined as metastases extending beyond the
regional nodes.
Statistical Analysis
Survival was measured from the date of randomization to the
date of death or most recent follow-up visit. Estimates of median
survival are based on the Kaplan–Meier method; group compar-
isons of survival involving individual variables were based on the
log-rank test. For categorical data, group comparisons were based
on the chi-square or Fisher’s exact test. Freedman’s method
23
was
used to estimate the sample size required to detect an improve-
ment in two-year survival of 20 percentage points after chemo-
therapy and radiotherapy over a base-line survival rate of 23 per-
cent for surgery alone — the survival rate for adenocarcinoma in
the facility at the commencement of the study. With an alpha er-
ror of 5 percent and a power of 80 percent, the number of
patients required in the study was estimated to be 190. Early
indications of a clinically relevant difference between treatments
suggested that an interim analysis should be undertaken. The trial
was closed six years after it began because a statistically significant
difference between the groups was found.
RESULTS
Demographic Data
Between May 1990 and September 1995, 113 pa-
tients with adenocarcinoma of the esophagus were
recruited. Fifty-eight patients were randomly as-
signed to receive chemotherapy and radiotherapy be-
fore surgery, and 55 were assigned to receive surgery
alone. The characteristics of the patients in the two
groups were similar before treatment (Table 1). While
the trial was in progress an additional 45 patients
The New England Journal of Medicine
Downloaded from nejm.org at ST LUKES INST CANCER RSCH on June 8, 2017. For personal use only. No other uses without permission.
Copyright © 1996 Massachusetts Medical Society. All rights reserved.
464
August 15, 1996
The New England Journal o
f
Medicine
underwent resection but did not undergo random-
ization for the following reasons: 14 were older than
76 years, 6 had distant metastases, 3 had uncertain
histologic findings, 2 had complete dysphagia, 2 had
previously undergone complex esophageal surgery,
1 required laryngectomy, 1 had recurrent tumor, and
10 chose surgery or chemoradiotherapy; in 6 the rea-
son for not undergoing randomization was not doc-
umented.
The median length of follow-up for all patients
was 10 months (range, 0.1 to 59). The median fol-
low-up for patients who died was 7.5 months (range,
0.1 to 37), whereas for patients who were still alive
as of the most recent follow-up visit it was 18
months (range, 1 to 59). The median follow-up was
10 months (range, 0.1 to 59) for patients assigned
to multimodal therapy and 8 months (range, 0.1 to
38) for patients assigned to surgery.
Protocol Violations
The chemoradiotherapy protocol was violated in
10 instances, and the surgery protocol in 1. Table 2
lists the protocol violations.
Treatment-Related Morbidity
Treatment-related toxicity was low, and the regi-
men was well tolerated. Six patients in the multimo-
dal group (10 percent) had grade III toxic reactions,
as defined by WHO criteria (three gastrointestinal,
two hematologic, and one cardiac); two patients had
grade IV toxic reactions (one cardiac and one gas-
trointestinal); and one patient had a fatal hemor-
rhage from the tumor bed during treatment.
Treatment was delayed in two patients because of
leukopenia, which resolved within two weeks. Two
patients whose ECOG performance status deteriorat-
ed did not undergo surgery and were included in the
group with grade III gastrointestinal toxic reactions.
Postoperative Complications
Respiratory complications occurred in 28 patients
in the multimodal group and 32 patients in the sur-
gery group. Fourteen cardiac events occurred in the
chemoradiotherapy group, as compared with 13 in
the surgery group. There were two anastomotic leaks,
one recurrent nerve palsy, and one chylothorax in
each group; one patient in the multimodal group
had severe postoperative pancreatitis, and one pa-
tient in the surgery group had a postoperative hem-
orrhage requiring a second operation.
Mortality during Hospitalization
Seven patients died in the hospital, for a 90-day
in-hospital mortality rate of 6 percent. Of the five
patients in the multimodal group who died during
hospitalization, three had completed the protocol.
One died of postoperative hemorrhage, one of chy-
lothorax, and one of an anastomotic leak. Two pa-
tients did not complete the multimodal protocol.
One died preoperatively of a hemorrhage from the
tumor bed, and one with complete dysphagia who
had surgery before completing chemoradiotherapy
died postoperatively of sepsis.
Two patients assigned to surgery died of sepsis in
the hospital. A chylothorax developed in one patient
within three weeks after surgery, and an iatrogenic
perforation occurred in the second after attempted
dilatation of the tumor.
Response to Chemoradiotherapy
At the time of surgery, only 42 percent (23 of 55)
of the patients in the multimodal group who could
be evaluated had positive lymph nodes or metasta-
ses, as compared with 82 percent (45 of 55) of the
patients in the surgery group (P
0.001) (Table 3).
A complete pathological response occurred in 13 of
the 52 patients in the multimodal group (25 per-
cent) who underwent resection, including 1 patient
who died and had no viable tumor at autopsy and
1 patient who only had high-grade dysplasia in the
tumor bed. In two patients overt metastases devel-
oped during treatment.
Survival
A comparison of the two treatment groups based
on the intention-to-treat principle showed a survival
advantage for the multimodal group (median surviv-
al, 16 months, as compared with 11 months in the
T
ABLE
1.
C
HARACTERISTICS
OF
THE
T
WO
T
REATMENT
G
ROUPS
AT
B
ASE
L
INE
.
C
HARACTERISTIC
S
URGERY
A
LONE
(N
55)
M
ULTIMODAL
T
HERAPY
(N
58)
Age (yr)
Median
Range
65
37–75
65
47–75
no. of patients
Sex
Male
Female
44
11
39
19
Tumor site
Middle third of esophagus
Lower third of esophagus
Cardia
5
27
23
11
31
16
Type of operation
Three stage 12 13
Lewis–Tanner 22 21
Abdomen and left side of
chest
17 15
Transhiatal 2 0
Abdominal 2 2
None 0 7
Barrett’s mucosa 21 22
The New England Journal of Medicine
Downloaded from nejm.org at ST LUKES INST CANCER RSCH on June 8, 2017. For personal use only. No other uses without permission.
Copyright © 1996 Massachusetts Medical Society. All rights reserved.
COMPARISON OF MULTIMODAL THERAPY AND SURGERY FOR ESOPHAGEAL ADENOCARCINOMA
Volume 335 Number 7
465
group treated with surgery alone; P
0.01) (Fig. 1).
When the two groups were compared on the basis
of the treatment actually received, the median sur-
vival was 32 months in the multimodal group and
11 months in the surgery group (P
0.001) (Fig.
2). When survival at one, two, and three years was
calculated on the basis of the intention to treat, 52,
37, and 32 percent of the patients assigned to multi-
modal therapy were alive, as compared with 44, 26,
and 6 percent of those assigned to surgery alone;
the difference reached statistical significance at three
years (P
0.01). The survival rates at three years ac-
cording to the treatment actually received were 37
percent (10 of 27 patients) in the multimodal group
and 7 percent (2 of 30) in the surgery group
(P
0.006) (Table 4).
Thirteen patients in the multimodal group had a
complete pathological response. One died during
treatment. A 72-year-old woman died five months
after surgery; there was no autopsy, but her family
physician attributed her death to tumor. The re-
maining 11 were alive and tumor-free 43, 36, 27, 21,
21, 16, 16, 5, 4, 4, and 2 months after resection.
Nineteen patients in the multimodal group had
residual disease confined to the esophagus. As of this
writing the median survival for this subgroup had
not been reached: 11 were alive, none with overt
disease, after a median follow-up of 37 months
(range, 10 to 59). Eight had died, five of proven and
one of suspected recurrence, after a median survival
of 8 months (range, 3 to 32). Thirteen had stage 3
disease, four of whom were alive at 54, 43, 18, and
4 months, whereas nine had died after a median of
7 months (range, 2 to 11), all but one of recurrent
disease.
Ten patients assigned to surgery had disease con-
fined to the esophagus and had a median survival of
14 months (range, 0.3 to 38). Three of these pa-
tients were alive at 25, 29, and 38 months, whereas
T
ABLE
2.
P
ROTOCOL
V
IOLATIONS
L
EADING
TO
THE
W
ITHDRAWAL
OF
11 P
ATIENTS
FROM
THE
S
TUDY
.
P
ATIENT
N
O
.
S
EX
/
A
GE
(
YR
)T
REATMENT
R
ECEIVED
C
OMMENT
F
OLLOW
-
UP
(
MO
)
Assigned to chemotherapy, radiotherapy, and surgery
1 F/75 No treatment Died of probable myocardial infarction
before treatment was initiated
0
2 M/63 Treatment interrupted Upper gastrointestinal hemorrhage during
treatment
11
3 F/70 Treatment interrupted ECOG performance status deteriorated
during treatment
5
4 M/68 Treatment interrupted Complete dysphagia developed 0
5 M/69 Treatment interrupted Pericarditis developed during treatment 4
6 M/73 Treatment completed Fatal bleeding from tumor bed; no tumor
at autopsy
0
7 F/75 Treatment completed ECOG performance status deteriorated
during treatment
3
8 M/40 Treatment completed Lung metastases developed 10
9 M/60 Treatment completed Myocardial infarction after treatment; no
surgery
29
10 M/74 Treatment completed Lung metastases developed 3
Assigned to surgery only
11 M/64 Emergency surgery Iatrogenic perforation, delayed referral 0.5
*One patient who completed the chemoradio-
therapy protocol died before surgery, and no viable
tumor was identified at autopsy. A second patient
who underwent resection had no invasive cancer but
had an area of high-grade dysplasia in the mucosa at
the site of the tumor.
T
ABLE
3.
P
ATHOLOGICAL
S
TAGE
OF
THE
T
UMOR
AT
THE
E
ND
OF
T
REATMENT
A
CTUALLY
R
ECEIVED
.
T
UMOR
STAGE
SURGERY ALONE
(N 55)
M
ULTIMODAL
THERAPY (N 58)
no. of patients
0 0 13*
121
2a 8 18
2b 2 7
33813
453
Unknown 0 3
The New England Journal of Medicine
Downloaded from nejm.org at ST LUKES INST CANCER RSCH on June 8, 2017. For personal use only. No other uses without permission.
Copyright © 1996 Massachusetts Medical Society. All rights reserved.
